The role covers the whole recommendation and matching stack, from models and real-time serving to data infrastructure, and you would own it from start to finish.
Responsibilities
Design and build retrieval, ranking and re-ranking systems that match ads to recipients.
Create embedding and vector search infrastructure that selects ads in real time.
Tune ranking to balance click-through rate, advertiser return, publisher revenue and subscriber relevance.
Run the feature store and serving path, covering offline training, online inference and pipeline reliability.
Use LLMs and foundation models for subscriber tagging, content understanding and user modeling.
Carry projects from prototype to production, including online learning, feedback loops, data pipelines and warehouse transformations.
Build dashboards and automated scorecards that track performance and key metrics.
Set priorities for the recommendation system with company and engineering leadership, with a path toward leading an ML team.
Requirements
You have 5 to 16 years of experience, with at least 3 years on recommendation, ranking or advertising systems.
You have shipped a production recommendation or matching system that covers candidate generation, ranking, embedding-based retrieval, serving and evaluation.
You have built ad ranking, click-through or conversion prediction, or ad matching models.
You have hands-on work with embeddings and real-time vector search tools such as FAISS, Milvus, Pinecone, pgvector or similar.
You have built data pipelines and transformation layers using SQL, dbt or similar tools, along with Postgres or columnar analytical databases.
You live in the NYC area, or in a US time zone near Eastern, and can travel to NYC quarterly.
You hold US work authorization and will not need visa sponsorship.
Early-stage startup or founder experience is helpful.
Advanced study in computer science or machine learning is helpful.